Motion Capture and 3D modelling

This research area is focused on the processing and utilization of motion capture system data (MOCAP) and three-dimensional scanning. The motion capture system allows monitoring of passive marker trajectories in 3D. Typically, markers are placed on moving objects (persons, robots, as well as animals, cars, flying objects etc.). Recordings may be processed in order to analyze movements of a tracked object to identify, for example, its behaviour or properties, and in order to reproduce the recorded (and often modified) data by animating a virtual 3D model that can either correspond to the original object, or be completely different.

The 3D scanning system uses structured light illumination of a scanned object and a camera to capture deformations of the projected structure caused by curves and different positions of individual sides of an object. Special software is then used to calculate a 3D surface model. The model can be further used to analyse three-dimensional properties of an object; it can also be preserved for possible model creation or its further use in the virtual environment (animations, etc.).

Our research is focused on the following areas:

Human Body and Face Animation

Typical tasks to illustrate the use of our motion-capture system include animations for gaming or film-making industries, where actions of a human actor dressed in a special suit with markers attached are captured. The movements hereby captured are edited and typically used for a visually different animated character (all types of fictional characters). As motion-capture data from a human actor (maintaining movement dynamics) are employed for the animation, the animation of a virtual character is realistic.

A special headset using four cameras to monitor 3D trajectories of markers applied or attached onto the face of a scanned person is used for facial animation. Easy application of markers by just a marker pen greatly simplifies the process of animation production.

However, the core of animation research is focused on designing realistic and natural systems for (audio)visual man-to-machine communication, namely research of audiovisual speech synthesis and sign-language synthesis aiming to create the most comprehensible, natural and realistic communication tool (lip animation of a speaking mouth, sign-language animations, etc.).

Analysis of Human Body Movements

Our research is namely focused on the fields of physiotherapy and orthopaedics – proper body posture, modelling of human body movements, such as walking, analysis of body movements during sports activities, etc. 

Motion Tracking of Rigid Mechanical Objects

One of the tasks in this field is motion tracking of rigid objects, typically in robotic applications. The objective is to track the actual position of a robot. Different types of objects are employed– from production line robots to flying drones (quadrocopters, octocopters, etc.). Actual position data are then used by a cooperating team to design robot control in such events when other sensors fail, e.g. because of GPS signal loss, slippery surface, etc.